ABSTRACT
In this work, a speech enhancement interface will be developed in order to be utilized for coding applications, synthesis and recognition. So, we suggest a single channel noise reduction in a noisy environment. The proposed approach combines the Continuous wavelet transform and the wiener filters with the aim of improving the speech intelligibility and quality. The proposed algorithm is tested in various scenarios and shows its robustness, in different noisy environments in term of SNR (signal-to-noise ratio), SSNR (segmental SNR), MSE (Mean Square Error), PESQ (perceptual evaluation of speech quality), Time Evolution and Spectrogram Analysis.
